Comparative Analysis of Expression Data Reveals Novel Targets of Senescence-Associated microRNAs

Marco Napolitano1, Marika Comegna2,3, Mariangela Succoio2,3, Eleonora Leggiero3, Lucio Pastore2,3, Raffaella Faraonio2,3, Filiberto Cimino1,2,3*, Fabiana Passaro2,3* 1 IRCCS SDN Foundation, Naples, Italy, 2 Department of Molecular Medicine and Medical Biotechnologies, University of Naples Federico II, Naples, Italy, 3 CEINGE – Advanced Biotechnologies, Naples, Italy

Abstract In the last decades, cellular senescence is viewed as a complex mechanism involved in different processes, ranging from tumor suppression to induction of age-related degenerative alterations. Senescence-inducing stimuli are myriad and, recently, we and others have demonstrated the role exerted by microRNAs in the induction and maintenance of senescence, by the identification of a subset of Senescence-Associated microRNAs (SAmiRs) up-regulated during replicative or stress- induced senescence and able to induce a premature senescent phenotype when over-expressed in human primary cells. With the intent to find novel direct targets of two specific SAmiRs, SAmiR-494 and -486-5p, and cellular pathways which they are involved in, we performed a comparative analysis of profiles available in literature to select down-regulated upon replicative senescence of human primary fibroblasts. Among them, we searched for SAmiR’s candidate targets by analyzing with different target prediction algorithms their 3’UTR for the presence of SAmiR-binding sites. The expression profiles of selected candidates have been validated on replicative and stress-induced senescence and the targeting of the 3’UTRs was assessed by luciferase assay. Results allowed us to identify Cell Division Cycle Associated 2 (CDCA2) and Inhibitor of DNA binding/differentiation type 4 (ID4) as novel targets of SAmiR-494 and SAmiR-486-5p, respectively. Furthermore, we demonstrated that the over-expression of CDCA2 in human primary fibroblasts was able to partially counteract etoposide-induced senescence by mitigating the activation of DNA Damage Response.

Citation: Napolitano M, Comegna M, Succoio M, Leggiero E, Pastore L, et al. (2014) Comparative Analysis of Gene Expression Data Reveals Novel Targets of Senescence-Associated microRNAs. PLoS ONE 9(6): e98669. doi:10.1371/journal.pone.0098669 Editor: Gianfranco Pintus, University of Sassari, Italy Received March 3, 2014; Accepted May 6, 2014; Published June 6, 2014 Copyright: ß 2014 Napolitano et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was supported by the Ministero dell’Universita` e della Ricerca Scientifica e Tecnologica (MIUR PRIN 2007, MIUR MERIT RBNE08HWLZ_004) and POR Campania FESR 2007-2013 Project BIOFRAME. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] (FC); [email protected] (FP)

Introduction Distinctive features of senescent cells include enlarged and flattened morphology, the appearance of senescence-associated Described for the first time in 1961 by Hayflick and Moorhead heterochromatin foci (SAHF), accumulation of senescence-associ- [1] as a process that limited the proliferation of normal human ated DNA-damage foci (SDFs) and expression of Senescence- cells in culture, cellular senescence currently refers to the Associated b-galactosidase (SA-b-gal) [4]. essentially irreversible growth arrest that occurs when proliferating Many senescence-inducing stimuli cause epigenomic disruption cells encounter a genotoxic stress and reveals a complex or genomic damage, like the gradual attrition of telomeres with phenomenon incorporating both genetic and environmental each S phase [7], that generates a persistent DNA damage components acting through convergent pathways. With the response (DDR), which initiates and maintains the senescence possible exception of embryonic stem cells [2], most division- growth arrest of human cells both in culture and in vivo [8–9]. competent cells, including adult stem cells and some tumor cells, Persistent DDR signaling generated at nontelomeric sites also can undergo senescence [3]. leads to the senescence growth arrest, such as that derived by Cellular senescence is thought to have evolved as a mechanism strong mitogenic signals delivered by certain or highly to prevent that damaged DNA could be replicated and passed on expressed pro-proliferative genes, that cause misfired replication to future generations of cells, thus being considered a tumor origins and replication fork collapse [10–12]. Treatments with suppressor mechanism [3]. Nevertheless, despite their inability to cytotoxic chemotherapeutic agents, such as etoposide, that cause replicate, senescent cells are metabolically active and develop an DNA double strand breaks, also induce premature senescence via aberrant gene expression profile with proinflammatory behaviour, the pathway [13]. Senescence can also occur, however, the so-called Senescence Associated Secretory Phenotype (SASP), without detectable DDR signaling. These stresses could include that can induce or accelerate changes in normal surrounding inappropriate substrates or serum or oxidative stress, as the tissues, explaining the possible implication in tumor promotion, Reactive Oxygen Species (ROS) production after treatment with aging and age-related pathologies [4]. Very recently, it has been oxidative stress agents, such as the glutathione depletor Diethyl- reported that cellular senescence contributes also to embryonic maleate (DEM) [14–15]. development, both in mice and humans [5–6].

PLOS ONE | www.plosone.org 1 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

In the last years microRNAs (miRs) have added a new layer of The housekeeping beta-actin (ACTB) mRNA was used as an complexity to the comprehension of molecular mechanisms internal reference gene for normalization. PCR reactions were underlying senescence. Each miR can recognize up to hundred performed on biological duplicates or triplicates and in experi- different targets, thus influencing a large variety of cellular mental triplicate. Fold changes were calculated using 22DDCt processes [16]. We and others have recently demonstrated that method, by the formula: 22(sample DCt - calibrator DCt), comparing some miRs are involved in the process of cellular senescence [17– results from experimental samples (Replicative senescent cells, 20]. In particular, we have found a subset of five Senescence- Etoposide-induced senescent cells or DEM-induced senescent Associated miRs (SAmiRs) up-regulated during Human Diploid cells) with both a calibrator (young PDL 33 cells for RS; DMSO Fibroblasts (HDFs) replicative senescence, whose ectopic expres- treated cells for EIS and DIS) and the reference gene ACTB. DCt sion in young cells promoted the premature senescence program is the difference between the amplification fluorescent thresholds by inducing DNA damage and ROS accumulation. of the gene of interest and ACTB. The list of the primers used is The main aim of this study has been the identification of novel reported in Table S2. mRNA targets of two selected SAmiRs, SAmiR-486-5p and TaqMan MiRNA Assay Kit (Applied Bio-systems, Foster City, SAmiR-494. We chose SAmiR-486-5p, as we previously showed CA) was used to detect the expression of mature miRNAs. Briefly, that this miR was the most robust up-regulated upon replicative 100 ng of total RNA was reversely transcribed (RT) at 16 uC for senescence and significantly up-regulated upon Etoposide-Induced 30 min, 42 uC for 30 min and 85 uC for 5 min in 15 ml reaction Senescence (EIS) of HDFs and in human primary skin fibroblasts volume. Two ml of RT product were used for PCR reaction in a from old donors [17]. Furthermore, Kim and colleagues demon- final volume of 20 ml. The PCR reaction started with an initial strated that the over-expression of SAmiR-486-5p was able to denaturation step at 90 uC for 10 min, followed by 40 cycles of 95 induce premature senescence also in human adipose tissue-derived uC for 15 sec and 60 uC for 1 min. Small nucleolar RNA RNU6 stem cells, by inhibiting SIRT1 [21]. We also chose SAmiR-494, (Applied Biosystems, Foster City, CA) was used for normalization. as its over-expression seemed to better recapitulate all the features PCR reactions were performed in triplicate and fold changes were of a senescent phenotype, including reduced cell proliferation, calculated using 22DDCt method, where DCt is the difference induction of SA-b-gal, SAHFs, DNA damage, SDFs and ROS between the amplification fluorescent thresholds of the miRNA of accumulation. Moreover, it resulted up-regulated also in DEM - interest and the RNA of RNU6. Induced Senescence (DIS) [17] and its over-expression was able to induce senescence not only in HDFs, but also in cancer cells [22]. Cell cultures, treatments and transfections Our strategy has been to look for new SAmiR targets among the Normal human primary fibroblasts IMR90 and human mRNAs down-regulated in HDF senescent cells, comparing six embryonic kidney HEK-293 cells were obtained from American different gene expression profiles available in literature, selecting Type Culture Collection (Manassas, VA). Cells were cultured in genes whose expression was reduced in at least three different Dulbecco’s modified Eagle’s medium (DMEM) supplemented with arrays and searching for SAmiR’s responsive elements into their 10% (v/v) fetal bovine serum and 1% penicillin/streptomycin 3’UTR by the use of different target prediction algorithms. (Gibco). Cultures were maintained at 37 uC in a 5% CO2- This approach allowed us to identify Cell Division Cycle humidified atmosphere. Associated 2 (CDCA2) and Inhibitor of DNA binding/differen- The IMR90 population doubling level (PDL) was calculated by tiation type 4 (ID4) as novel targets of SAmiR-494 and SAmiR- using the formula: DPDL = log(nh/ni)/log2, where ni is the initial 486-5p, respectively. Moreover, we have also demonstrated that number of cells and nh is the final number of cells at each passage. the transient over-expression of CDCA2 in HDFs is able to The cells were used at 33 PDL (young) or 58 PDL (senescent) partially counteract the premature senescent phenotype induced (Fig.S1). by etoposide treatment. To induce premature senescence, IMR90 at PDL 33 were treated with 150 mM DEM on alternate days for 10 days or with Materials and Methods 20 mM etoposide (both from SIGMA-ALDRICH) for 24 h and then subcultured for 10 days more (Fig.S1). Bioinformatics analysis Transfection of IMR90 cells at PDL 33 with synthetic pre-miR We took advantage of available data reporting gene expression precursors (Ambion), miR inhibitors (Exiqon) or siRNAs (Dhar- profiles in replicative senescent HDFs. Normalized data from 6 macon) were performed using Lipofectamine RNAiMAX Trans- microarrays [23–28] have been crossed in order to select a list of fection Reagent (Life Technologies) with the reverse protocol common genes down-regulated in senescence, with a fold following the manufacturer’s instructions. Pre-miRs, miR inhib- variation$1.5. We obtained a list of 139 genes down-regulated itors and siRNAs were used at 100 nM. upon HDFs replicative senescence (Table S1). Functional anno- Transfection of IMR90 cells at PDL 33 with CMV-CDCA2 tations were obtained on PubMed.gov. For the identification of and CMV-ID4, as well as co-transfection of HEK-293 with both putative targets of SAmiR-494 or SAmiR-486-5p, the genes in the luciferase constructs and pre-miRs were performed using Lipo- list have been analyzed with four different target prediction fectamine 2000 (Life Technologies) following the manufacturer’s algorithms (Target Scan v6.2, miRDB, Diana, miRanda) and instructions. putative targets predicted by at least two algorithms have been selected for further studies. Plasmid construction With the exception of ID4, for which it has been cloned a Real time PCR portion of 300 bp containing the putative SAmiR-486-5p binding Total RNA was extracted with TRIzol Reagent (Life Technol- site, the whole 3’UTRs sequences of CDCA2, FOXM1, NUSAP1 ogies) and quantified by Nanodrop (Thermo Scientific, Wilming- and BUB1b have been cloned by PCR amplification on human ton, DE). The first-strand cDNA was synthesized according to the genomic DNA, using primer pairs with XhoI and SalI restriction manufacturer’s instructions (SS VILO Mastermix- Life Technol- enzyme sites in the forward primers and XbaI in the reverse ogies). Real-time RT-PCR was carried out on an iCycler (BioRad) primers. A 430 bp portion of the 3’UTR of human OLFM4, using Express Greener QPCR Master mix (Life Technologies). containing the validated SAmiR-486-5p binding site [29], was

PLOS ONE | www.plosone.org 2 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets cloned as positive control using primer pairs with SalI restriction Immunofluorescence enzyme sites in the forward primers and XbaI in the reverse To perform c-H2AX staining, IMR90 at PDL33 were plated on primers. All PCR products were cloned into the pMIR-GLO glass coverslips, transfected with CMV-NEO control vector or vector (Promega) between the XhoI and the XbaI site, CMV-CDCA2 for 24 h and, then, treated with 20 mM etoposide. downstream the Firefly luciferase (Normal clones). The inverted Coverslips were collected at 0 h, 6 h and 18 h after treatment, 3’UTR of CDCA2, ID4 and OLFM4 (Reverse clones) were fixed with 4% paraformaldehyde in PBS for 30 minutes at RT, cloned by digestion of Forward clones with SalI, that allowed the permeabilized with 10% FBS, 1% BSA, 0,2% Triton in PBS for excision of the 3’UTR, and by recloning digested fragment in SalI 15 minutes at RT and incubated with Anti-phospho-Histone unique site. The orientation of the inserted fragments were H2A.X Ser 139 primary antibody (Millipore) for 2 h at RT. After established by digestions and confirmed with sequencing. The 4 washes of 5 minutes each, coverslips were incubated with an 3’UTRs of CDCA2 and ID4 containing point mutations in the Alexa-488 goat anti-mouse antibody (Life Technologies) for 1 h at SAmiR seed region (Mutated clones) were obtained by PCR using RT, counterstained with DAPI and mounted in Moviol on glass the Quik Change II XL site direct mutagenesis kit (Agilent), slides. Samples were observed with an epifluorescent microscope following the manufacturer’s instructions. Mutations were con- Leica DM IL LED FLUO (Leica Microsystems) and at least 300 firmed by sequencing. cells were counted in triplicate experiments. The coding sequences of CDCA2 and ID4 were amplified by PCR from ULTIMATEHORF CLONE ID IOH44066 and Western blotting ULTIMATEHORF CLONE ID IOH12413 (Life Technologies), IMR90 cells were harvested following washing with PBS. Cells respectively, and cloned into the pEGFPN1 vector, in place of the were lysed in a buffer containing 0.02M HEPES (pH 7.9), 0.4M GFP coding sequence, between the AgeI and NotI sites. The NaCl, 0.1% NP-40, 10% (v/v) glycerol, 1 mM NaF, 1 mM obtained vectors were named CMV-CDCA2 and CMV-ID4. All sodium orthovanadate and a protease inhibitory cocktail (Sigma primers used for plasmids construction and the oligos containing Chemical Co. St. Louis, MO). Extracts were subjected to Sodium the mutated SAmiR seed regions are reported in Table S3. Dodecyl Sulfate (SDS)-polyacrylamide gel electrophoresis, fol- lowed by blotting to PVDF. The blots were probed with antibodies Luciferase Reporter Assay from Santa Cruz to human CDCA2, ID4, p53, p21, b-actin and For luciferase assays, human HEK-293 cells were plated at a-tubulin; from Cell Signaling to human phospho-ATM (Ser 86104 cells per well on 48 well plates (BD Falcon) 12 h before 1981) and phospho-p53 (Ser 15) ; from Sigma-Aldrich to human transfection. The Normal, Reverse or Mutated luciferase con- ATM. structs (100 ng) were co-transfected with 50 nM pre-miRs. All transfection experiments were done in triplicate and each Statistical analysis experiment was repeated three times. The Renilla luciferase Statistical analysis were carried out using the Student’s t test and reporter, contained into the pMIR-GLO vector, was used as an data were considered significant at a value of p,0.05. internal control. The luciferase activity was measured 48 hours after transfection using a Dual Luciferase Reporter Assay System Results n (Promega) according to manufacturer’s instructions, on a 20/20 A set of mRNAs is down-regulated in human senescent Luminometer instrument (Turner BioSystems). The data gener- fibroblasts ated were expressed as relative to control-miR transfected cells, To select candidate targets of SAmiR-494 or SAmiR-486-5p, after normalization to Renilla luciferase reading. we speculated that SAmiRs induced upon Replicative Senescence (RS) of HDFs could contribute to the suppression of genes that SA-b-gal assay and BrdU assay must be kept down-regulated on the induction of RS. Thus, we SA-b-gal was assayed according to Dimri et al. [30]. Briefly, cells generated a list of mRNAs down-regulated on HDFs senescence were washed twice with PBS, fixed with 2% formaldehyde and by comparing the normalized data of six different microarray gene 0.2% glutaraldehyde in PBS, and washed twice in PBS. Then, cells expression profiles available on public databases [23–28]. This were stained overnight in X-gal staining solution [1 mg/ml X-gal, analysis allowed to select 139 mRNAs down-regulated($1.5 folds) 40 mM citric acid/sodium phosphate (pH 6.0), 5 mM potassium in at least 3 out of 6 different arrays (Table S1). As summarized in ferricyanide, 5 mM potassium ferrocyanide, 150 mM NaCl, Fig. 1A, we analyzed the 139 mRNAs for the presence of 2 mM MgCl2]at37uC. consensus motifs for SAmiR-494 and/or for SAmiR-486-5p, by For BrdU (5-bromo-2-deoxyuridine) incorporation assay using four different target prediction algorithms (Target Scan v6.2, (ROCHE), cells were seeded on glass coverslips and transfected miRDB, Diana, miRanda) and focusing on the results common to with siRNAs to induce the knock-down of SAmiR targets, or at least two algorithms. This screening led to the generation of the transfected with CMV-CDCA2, CMV-ID4 or CMV-NEO (as list of candidate targets shown in Fig. 1B, with 20 putative targets control plasmid) and treated with 20 mM etoposide to induce EIS of SAmiR-494, 7 of SAmiR-486-5p and 3 common to both or with 150 mM DEM to induce DIS. 72 h after transfection SAmiRs (in grey). Among them, there are many mRNAs encoding (siRNAs) or 24 h after treatments (48 h after plasmids over- involved in regulation (e.g. CCNE2, NUSAP1, expression), cells were incubated for 4 h with BrdU (10 mM) and ZWINT) and DDR (e.g. RAD51, RAD51AP1, DEK), two fixed following the kit instructions. Coverslips were incubated with biological processes modified by ectopic expression of SAmiRs a primary anti-BrdU and a secondary fluorescein-conjugated [17] (see also Table S1 for functional annotations). Interestingly, antibodies and then counterstained with Hoechst 33258, rinsed some of the candidates are members of the same family and mounted in Moviol on glass slides. The fluorescent signal was (e.g. BUB3 and BUB1b; CDCA2, CDCA4 and CDCA7; RFC2 visualized with an epifluorescent microscope Leica DM IL LED and RFC3). We excluded BIRC5 (survivin) from further FLUO (Leica Microsystems). At least 300 cells were counted in investigation, as it was already validated by others [31]. triplicate experiments. To validate the results of our comparative analysis, we investigated in IMR90 cells the expression profile of putative

PLOS ONE | www.plosone.org 3 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

Figure 1. Strategy to identify putative targets of SAmiRs. A) The 3’UTR sequences of the 139 mRNAs down-regulated in replicative senescent fibroblasts and reported in Table S1 were analyzed with four different target prediction algorithms. This in silico analysis revealed 30 putative SAmiR targets: 20 of SAmiR-494, 7 of SAmiR-486-5p and 3 common to both SAmiRs. B) List of the 30 predicted target genes of both SAmiR-494 or SAmiR- 486-5p. In grey, the three putative targets common to both SAmiRs. doi:10.1371/journal.pone.0098669.g001

PLOS ONE | www.plosone.org 4 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets targets upon induction of RS, EIS and DIS (Fig. 2). With the luciferase expression compared to unrelated pre-miR transfected exception of CDCA7, SOCS2 and ZNF367, whose expression in cells (Fig. S3A). IMR90 was undetectable (not shown), the results obtained by Real To further characterize the functionality of predicted target sites Time PCR demonstrated a common signature of gene expression in the 3’UTRs of CDCA2 and ID4, the corresponding reverse in replicative and stress-induced senescence, with 17 out of 26 fragments or the fragments mutated at the seed region of putative putative target genes that resulted significantly (p,0.05) down- target sites were also generated. As shown in Fig. 3B, the luc- regulated, with a fold variation$2, upon RS, 14 out of 26 upon reverse constructs (R), as well as the constructs bearing mutated EIS and 7 out of 26 upon DIS. These results prompted us to select UTRs (M) were unaffected by the cognate SAmiR transfection (see for further investigation the 7 candidates, highlighted in Fig. 2 by a also Fig. 3C for point mutations into SAmiRs seed regions), thus star, whose expression resulted down-regulated in all the examined strongly suggesting that CDCA2 and ID4 were direct targets of conditions (RS, EIS and DIS). selected SAmiRs. We also investigated the expression profiles of the two targets by A subset of putative targets are down-regulated upon Western blot analysis, upon SAmiRs over-expression or down- SAmiR over-expression in human fibroblasts regulation in PDL 33 IMR90 cells. As shown in Fig. 3D and 3E, In order to identify direct targets of SAmiRs, considering that the decrease in CDCA2 and ID4 endogenous expression levels mRNAs targeted by a miR are generally degraded [32–33], we was well detectable at protein level. Noteworthy, the transfection analyzed by Real Time PCR the mRNA levels of BUB1b, of miR inhibitors caused the increase of target expression levels. CDCA2, FOXM1, ID4, MKI67, NUSAP1 and PCOLCE at day All together, these data demonstrated that CDCA2 and ID4 are 2 after the transfection of the cognate synthetic SAmiR precursor direct targets of SAmiRs-494 and SAmiR-486-5p, respectively. (pre-miRs). As shown in Fig. 3A, CDCA2, FOXM1 and NUSAP1 resulted significantly (p,0.01) down-regulated 2 days after Knock-down of CDCA2 or ID4 in young human primary SAmiR-494 pre-miR transfection, whereas, among SAmiR-486- fibroblasts does not induce senescence 5p putative targets, ID4 and, at a lower extent, BUB1b, showed a To analyze the role of CDCA2 and ID4 in senescence, we asked significant reduction upon pre-miR transfection. whether their knock-down by RNAi in young cells was able to induce premature senescence, as the up-regulation of the cognate CDCA2 and ID4 are direct target of SAmiR-494 and SAmiRs does. To this aim, we transfected siRNAs designed to SAmiR-486-5p, respectively silence CDCA2 or ID4 in young IMR90 at PDL 33 individually or To address whether CDCA2, FOXM1, NUSAP1, ID4 and as a mixture (Fig. S3B), and then we analyzed the cells until ten BUB1b, whose mRNAs resulted reduced after SAmiR over- days after transfection, in order to detect any signs of senescence, expression, were direct targets, luciferase constructs containing as the decreasing of cell proliferation by BrdU incorporation, the their 39 UTR sequences were generated. As shown in Fig. 3B, in change in cell morphology or the appearance of SA-b-gal. As the case of CDCA2 and ID4 the reporter gene expression was showed in Fig. 4A, the knock-down of these genes seemed to be significantly reduced by SAmiR-494 or SAmiR-486-5p pre-miR unable to affect cell proliferation, although a weak but significant transfection, respectively, with a variation of relative luciferase decrease in BrdU incorporation was detected in siCDCA2 cells. expression similar to the positive control OLFM4 (N). In contrast, Accordingly, knock-down cells did not senesce prematurely, as the co-transfection of FOXM1, NUSAP1 or BUB1b luciferase demonstrated by the absence of SA-b-gal staining 10 days after constructs with the cognate SAmiR didn’t show any change in siRNA transfection (Fig. 4B). Probably, neither the transient

Figure 2. Expression profile of putative targets upon the induction of replicative or stress-induced senescence. The expression levels of the 26 candidates were measured by Real Time PCR in replicative (RS: IMR90 cells at PDL 58), etoposide- (EIS: PDL 33 IMR90 cells treated with 20 mM etoposide for 24 h and then subcultivated for additional 10 days) and DEM-induced (DIS: PDL 33 IMR90 cells treated with 150 mM DEM on alternate days for 10 days) senescent cells. The mRNA relative expression was calculated by assigning the arbitrary value 1 to the amount found in young or DMSO-treated cells. SD is used to refer to the values obtained in 2 different experiments. Results showed that 7 mRNAs, highlighted by a star, resulted significantly (p,0.05) down-regulated, with a cut-off$2 folds, in all conditions. doi:10.1371/journal.pone.0098669.g002

PLOS ONE | www.plosone.org 5 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

Figure 3. Expression profile of putative targets upon SAmiRs ectopic expression and validation of CDCA2 and ID4. A) Expression levels of SAmiR-494 and SAmiR-486-5p putative targets 2 days after the transfection of the cognate SAmiR pre-miR in PDL 33 IMR90 cells. mRNA levels were measured by Real Time PCR and mRNA relative expression was calculated by assigning the arbitrary value 1 to the amount found in control cells transfected with a scramble pre-miR. SD refers to the values obtained in 3 different experiments and the difference was significant (** p, 0.01; * p,0.05). The quantification of the expression levels of SAmiRs after their ectopic over-expression is reported in Panel A of Figure S2. B) Luciferase constructs bearing the normal 3’UTRs (N), reverse 3’UTRs (R) or mutated 3’UTRs (M) of CDCA2 and ID4 were transfected in HEK293 cells together with the cognate pre-miR or control pre-miR. The normal and reverse 3’UTR of OLFM4 were used as positive control. Luciferase levels were reported as fold changes compared to the values measured in control pre-miR transfected cells, after normalization with Renilla luciferase activity. SD refers to the values obtained in 3 different experiments (* p,0.01). C) Wild type seed regions of SAmiR-494 and SAmiR-486-5p, respectively present into the 3’UTRs of CDCA2 and ID4, compared to the mutated seed regions used for luciferase assays. D) Western blot analysis of CDCA2 in IMR90 cells over-expressing SAmiR-494 (pre-miR) or a specific SAmiR-494 inhibitor (anti-miR). E) Western blot analysis of ID4 in IMR90 cells over-expressing SAmiR-486-5p (pre-miR) or a specific SAmiR-486-5p inhibitor (anti-miR). In both cases, the proteins resulted suppressed by the SAmiR over- expression, as well as they resulted up-regulated by the anti-miR transfection, if compared to the control scramble transfected cells. The quantification of the expression levels of SAmiRs after their ectopic over-expression in D and E is reported in Panel B of Figure S2. doi:10.1371/journal.pone.0098669.g003 down-regulation of individual SAmiR target, nor the knock-down Adoptive expression of CDCA2 promotes cell cycle of both targets have the ‘‘strength’’ to switching-on the senescence progression in EIS program, as the contemporaneous reduction of many targets We transfected CDCA2 or ID4 in young PDL 33 IMR90 cells, exerted by SAmiR’s over-expression do. one day before the induction of EIS or DIS. Then, we monitored Nevertheless, the down-regulation of CDCA2 or ID4 caused by the progression of senescence by checking for BrdU incorporation, the up-regulation of cognate SAmiRs could contribute to the cell morphology changes and appearance of SA-b-gal. acquisition of the final senescence phenotype. Thus, we investi- While DIS program seemed to be unaffected, the transient over- gated the possible role of CDCA2 or ID4 in cellular senescence by expression of CDCA2, but not ID4, was able to counteract the EIS determining the effects of their over-expression. program, promoting cell cycle progression (Fig. 5A), despite the DNA damage induced by etoposide (Fig. S4A). However, this is a

PLOS ONE | www.plosone.org 6 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

Figure 4. Knock-down of CDCA2, ID4 or both does not induce premature senescence in PDL 33 IMR90 cells. A) siRNAs designed to target the coding sequence of CDCA2 or ID4 were transfected, individually or as a mix, in PDL 33 IMR90 cells to knock-down the expression levels of target genes. After 72 h, transfected cells were incubated with BrdU for 4 h, then coverslips were fixed, incubated with a primary anti-BrdU antibody, washed and incubated with a secondary fluorescein-conjugated antibody, counterstained with Hoechst-33258 and counted by immunofluorescence. Counts of at least 1,000 cells were averaged and expressed as fold changes6SD, with respect to scrambled transfected cells (*** p,0.001). B) siRNA transfected cells were subcultivated for 10 days and stained for SA-b-gal. Counts of at least 300 cells were averaged and expressed as fold changes6SD, with respect to scrambled transfected cells. doi:10.1371/journal.pone.0098669.g004 temporary effect, as CDCA2 over-expressing cells finally arrested activation induced by etoposide treatment by recruiting PP1c their growth and showed SA-b-gal accumulation, just like control phosphatase to chromatin at damaged sites and causing the cells (Fig. S4B). dephosphorylation of activated ATM [34]. This phenomenon Etoposide provokes DNA strand breaks that induce senescence could also explain the resistance to EIS showed by IMR90 cells through the activation of the p53 pathway [13]. Furthermore, it over-expressing CDCA2. Therefore, we measured ATM activa- has been demonstrated that, in human non-tumorigenic cells tion in CDCA2 over-expressing cells 24 h after EIS induction. (MCF10A), the over-expression of CDCA2 attenuates DDR

Figure 5. Adoptive expression of CDCA2 promotes cell cycle progression in Etoposide-Induced Senescence. A) CDCA2 and ID4 coding sequences were transfected in PDL 33 IMR90 cells. Cells transfected with CMV-NEO plasmid were used as control. After 24 h, transfected cells were treated with 20 mM etoposide or 150 mM DEM for 24 h and then incubated with BrdU for 4 h. Coverslips were then fixed, incubated with primary anti-BrdU and secondary fluorescein-conjugated antibodies, counterstained with Hoechst-33258 and counted by immunofluorescence. Counts of at least 800 cells were averaged and expressed as fold changes6SD, with respect to control transfected cells (***p,0.01). B) and C) PDL 33 IMR90 cells were transfected with the coding sequence of CDCA2. After 24 h, transfected cells were treated with 20 mM etoposide for 24 h and then were collected to obtain protein extracts. Cell extracts from CMV-neo over-expressing cells served as control. Western Blot analysis was used to detect the levels of phosphorylated ATM (p-ATM Ser1981), ATM, phosphorylated p53 (p-p53 Ser15), p53, p21Cip1 and CDCA2 in the cell lysates. b-actin was used as a loading control. doi:10.1371/journal.pone.0098669.g005

PLOS ONE | www.plosone.org 7 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

As shown in Fig. 5B and 5C, the over-expression of CDCA2 is performs different regulatory functions mainly during embryo- accompanied by a reduction of ATM activation. This results in a genesis, being involved in the neural stem cell, oligodendrocyte reduced activation of p53 and a consequent weak expression of the and astrocyte differentiation. cyclin-dependent kinase inhibitor p21Cip1, which can explain the Like CDCA2, ID4 is expressed in several tumors where it seems sustained cell proliferation observed despite etoposide treatment. to play a key role in cellular transformation, immortalization, All these data demonstrate that also in primary human cells invasion, and in the metastatic process [41–42]. Thus, a decrease CDCA2 over-expression is able to antagonize the activation of in ID4 expression levels could be one of the mechanisms adopted ATM-dependent signal transduction modulating DDR sensitivity, by cellular senescence to antagonize the tumor development in thus preventing the arrest of cell-cycle progression of premature pre-cancerous cells. Nevertheless, the expression of ID4 is senescence via decreased expression of CDKIs. epigenetically silenced in prostate cancer and, very recently, it has been demonstrated that ectopic over-expression of ID4 Discussion promotes cellular senescence in prostate cancer cell line DU145 by increasing the expression of p16, p21, p27, E-cadherin and In this study, with the intent to find novel direct targets of two vimentin, but down-regulating expression. ID4 also poten- specific SAmiRs previously associated to the induction and tiated the effect of doxorubicin induced senescence and apoptosis maintenance of cellular senescence [17], we took advantage of [43]. These findings suggest that the role of ID4 in cellular gene expression data available in literature to select by a senescence could be strictly dependent on tissue or cell types. comparative analysis and validate a subset of genes, whose In conclusion, we have identified CDCA2 and ID4 as new expression was strongly reduced in replicative and stress-induced direct targets of two different miRs, whose up-regulation in HDFs senescent cells. Among these genes we found that CDCA2, a is correlated to the induction of cellular senescence. As expected, specific nuclear regulatory subunit of protein phosphatase 1 c both targets are down-regulated during replicative- or stress- (PP1c), and ID4, a member of a family of helix-loop-helix induced senescence, but cannot induce senescence when silenced factors, are direct targets of SAmiR-494 and SAmiR- in young HDFs. Nevertheless, for CDCA2, our results are in 486-5p, respectively. accordance with previous findings in pre-malignant (not cancer- In our study, the down-regulation in HDFs of CDCA2, ID4 or ous) or tumor cells, in which the levels of CDCA2 determine the both fails to cause a massive cell cycle arrest typical of premature activation threshold of the DNA damage checkpoint. Thus, our senescence, that instead occurs after the over-expression of their results indicate that CDCA2 sets the threshold for checkpoint negative regulators, SAmiR-494 or SAmiR-486-5p. Nevertheless, activation also in HDFs. the ectopic expression of CDCA2, but not of ID4, was able to partially counteract the progression of EIS by the reduction of ATM activation, thus avoiding the cell cycle arrest caused by the Supporting Information activation of p53 and of DNA damage checkpoints after etoposide Figure S1 Characterization of senescent IMR90 cells. A) treatment. This would make cells less sensitive to DNA damage. Characterization of replicative senescent IMR90 cells. Primary Thus, it can be speculated that the down-regulation of CDCA2 human fibroblasts IMR90 were grown in DMEM supplemented expression that occurs in HDFs during cellular senescence could with 10% (v/v) fetal bovine serum and 1% penicillin/streptomycin contribute to the accumulation of DNA damage that, in turn, for some months, until PDL 58. B) Characterization of cellular sustains, rather than provokes, the senescence program. senescence induced by etoposide treatment. PDL 33 IMR90 cells CDCA2, also known as Repo-Man, is involved in cell cycle were exposed to etoposide 20 mM for 24 h and then cultured for regulation [35–38), as well as in PP1c-dependent essential DDR 10 days before harvesting. C) Characterization of cellular regulation [34]. Our data are in accordance with previous findings senescence induced by DEM treatment. PDL 33 IMR90 cells from Peng and colleagues, reporting that CDCA2 recruits PP1c to were chronically exposed to DEM 150 mM on alternate days for chromatin to antagonize activation of ATM-dependent signal 10 days before harvesting. In all cases, cellular senescence was transduction in pre-malignant (not cancerous) cells [34]. assessed by SA-b-gal staining and gene expression profile [4]. For Moreover, CDCA2-dependent DDR regulation is strengthened SA-b-gal staining, representative images of control and senescent by CDCA2 over-expression during cancer progression, resulting in cells are reported. For each experiment, at least 300 cells were reduced DDR sensitivity. CDCA2 is frequently over-expressed in counted. For gene expression profiling, quantitative Real-Time many tumor cells, as neuroblastoma, melanoma, breast cancer and PCR analysis of cyclin A (CyclA), thymidylate synthase (THY), in oral squamous cell carcinoma [39]. In particular, experiments cyclin-selective ubiquitin carrier protein (UBI), interleukin-6 (IL6), on oral cancer cell lines showed that suppression of CDCA2 cyclin-dependent kinase inhibitors p21Cip1 (p21) and p16INK4A expression with shRNA significantly inhibits cellular proliferation, (p16) mRNAs are showed. Results are the mean of triplicate by arresting cell-cycle progression at the G1 phase through determinations. activation of the DDR in vitro, thus suggesting that up-regulation of (TIFF) CDCA2 in tumor cells might prevent the arrest of cell-cycle progression, via decreased expression of CDKIs and regulation of Figure S2 Quantification of SAmiR expression levels the DDR. after their ectopic over-expression. The expression levels of However, in our study CDCA2 over-expressing cells finally SAmiR-494 and SAmiR-486-5p were measured by Real Time encounter a senescent growth arrest. This behavior might be PCR in PDL 33 IMR90 cells transfected with 100 nM pre-miR. explained by the fact that the reduction of p53 and p21 activation The microRNA relative expression was calculated by assigning the is only partial. Other mechanisms, parallel to ATM activation, arbitrary value 1 to the amount found in control pre-miR may be involved in supporting the phosphorylation of p53. transfected cells. SD is used to refer to the values obtained in 3 Concerning ID4, it is an helix-loop-helix different experiments. In all cases, the difference was significant (p that, having lost the basic DNA-binding domain, acts as , 0.01). A) Data refers to SAmiR’s over-expression of Figure 3A; dominant-negative regulator by forming inactive heterodimeric B) data refers to SAmiR’s over-expression of Figure 3D and 3E. complexes with other helix-loop-helix transcription factors [40]. It (TIFF)

PLOS ONE | www.plosone.org 8 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

Figure S3 A) Luciferase assay of unvalidated SAmiR over-expressing CDCA2 induces premature senescence. predicted target genes. Luciferase constructs bearing the PDL33 IMR90 cells were transfected with control vector or a normal 3’UTRs of FOXM1, NUSAP1 (predicted targets of vector containing the coding sequence of human CDCA2 gene. SAmiR-494) and BUB1b (predicted target of SAmiR-486-5p) were After 24 h, transfected cells were treated with 20 mM etoposide for transfected in HEK293 cells, together with the specific SAmiR 24 h and then were subcultivated for 10 days before harvesting. pre-miR, control pre-miR or an unrelated pre-miR. Luciferase Cellular senescence was assessed by SA-b-gal staining. At least 300 levels were reported as fold changes compared to the values cells were counted. Representative images of control and senescent measured in control pre-miR transfected cells, after normalization cells are showed. with Renilla luciferase activity. B) SAmiR’s target knock- (TIFF) down by siRNAs. The expression levels of CDCA2 and ID4 were measured by Real Time PCR in PDL 33 IMR90 cells Table S1 Genes whose expression is down-regulated transfected with 100 nM siRNAs, individually or as a mix. The during senescence of human diploid fibroblasts. mRNA relative expression was calculated by assigning the (PDF) arbitrary value 1 to the amount found in scramble-transfected Table S2 Primer pairs for Real Time PCR. cells. SD is used to refer to the values obtained in 3 different (PDF) experiments. In all cases, the difference was significant (p , 0.05). (TIFF) Table S3 Primer pairs for plasmid construction. (PDF) Figure S4 A) Etoposide treatment of IMR90 cells over- expressing CDCA2 induces cH2AX foci. PDL 33 IMR90 Acknowledgments cells were transfected with control CMV-NEO vector or with CMV-CDCA2. After 24 h, transfected cells were treated with We thank Prof. Tommaso Russo for critical reading of the manuscript and 20 mM etoposide. Cells were fixed and examined by immunoflu- for helpful discussions. orescence for a-H2AX phosphorylated on Ser139 (cH2AX) at 0, 6 or 18 hours after treatment. Coverslips were washed and Author Contributions incubated with Alexa-488 Goat anti-rabbit antibody and coun- Conceived and designed the experiments: MN FP FC. Performed the terstained with DAPI. Counts of at least 300 cells were averaged experiments: MN MC MS EL FP. Analyzed the data: MN FP LP RF FC. and expressed as percent of cells positive to the presence of Wrote the paper: FP FC. cH2AX foci 6 SD. B) Etoposide treatment of IMR90 cells

References 1. Hayflick L, Moorhead PS (1961) The serial cultivation of human diploid cell 17. Faraonio R, Salerno P, Passaro F, Sedia C, Iaccio A, et al. (2012) A set of strains. Exp Cell Res 25: 585–621. miRNAs participates in the cellular senescence program in human diploid 2. Miura T, Mattson MP, Rao MS (2004) Cellular lifespan and senescence fibroblasts. Cell Death Differ 19 (4): 713–721. signaling in embryonic stem cells. Aging Cell 3: 333–343. 18. Gorospe M, Abdelmohsen K (2011) MicroRegulators come of age in senescence. 3. Campisi J, d’Adda di Fagagna F (2007) Cellular senescence: when bad things Trends Genet 27: 233–41. happen to good cells. Nat Rev Mol Cell Biol 8: 729–740. 19. Mancini M, Saintigny G, Mahe´ C, Annicchiarico-Petruzzelli M, Melino G, et al. 4. Rodier F, Campisi J (2011) Four faces of cellular senescence. J Cell Biol 192 (4): (2012) MicroRNA-152 and -181a participate in human dermal fibroblasts 547–556. senescence acting on cell adhesion and remodeling of the extra-cellular matrix. 5. Mun˜os-Espı´nD,Can˜amero M, Maraver A, Go´mez-Lo´pez G, Contreras J, et al. Aging (Albany) 4: 843–53. (2013) Programmed cell senescence during mammalian embryonic develop- 20. Smith-Vikos T, Slack FJ (2012) MicroRNAs and their roles in aging. J Cell Sci ment. Cell 155 (5): 1104–1118. 125: 7–17. 6. Storer M, Mas A, Robert-Moreno A, Pecoraro M, Ortells MC, et al. (2013) 21. Kim YJ, Hwang SH, Lee SY, Shin KK, Cho HH, et al. (2012) miR-486-5p Senescence is a developmental mechanism that contributes to embryonic growth induces replicative senescence of human adipose tissue-derived mesenchymal and patterning. Cell 155 (5): 1119–1130. stem cells and its expression is controlled by high glucose. Stem Cells Dev 21: 7. Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu CP, et al. (1998) Extension 1749–1760. of life-span by introduction of telomerase into normal human cells. Science 279: 22. Ohdaira H, Sekiguchi M, Miyata K, Yoshida K (2012) MicroRNA-494 349–352. suppresses cell proliferation and induces senescence in A549 lung cancer cells. 8. Rodier F, Coppe JP, Patil CK, Hoeijmakers WAM, Munoz DP, et al. (2009) Cell Prolif 45 (1): 32–38. Persistent DNA damage signaling triggers senescence-associated inflammatory 23. Binet R, Ythier D, Robles AI, Collado M, Larrieu D, et al. (2009) WNT16B is a cytokine secretion. Nat Cell Biol 11: 973–979. new marker of cellular senescence that regulates p53 activity and the 9. d’Adda di Fagagna F, Reaper PM, Clay-Farrace L, Fiegler H, Carr P, et al. phosphoinositide 3-kinase/AKT pathway. Cancer Res 69 (24): 9183–91. (2003) A DNA damage checkpoint response in telomere-initiated senescence. 24. Zhang H, Pan KH, Cohen SN (2003) Senescence-specific gene expression Nature 426: 194–198. fingerprints reveal cell-type-dependent physical clustering of up-regulated 10. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW (1997) Oncogenic ras chromosomal loci. Proc Natl Acad Sci U S A 100 (6): 3251–6. provokes premature cell senescence associated with accumulation of p53 and 25. Hardy K, Mansfield L, Mackay A, Benvenuti S, Ismail S, et al. (2005) p16INK4a. Cell 88: 593–602. Transcriptional networks and cellular senescence in human mammary 11. Lin AW, Barradas M, Stone JC, van Aelst L, Serrano M, et al. (1998) Premature fibroblasts. Mol Biol Cell 16 (2): 943–53. senescence involving p53 and p16 is activated in response to constitutive MEK/ 26. Zhang H, Herbert BS, Pan KH, Shay JW, Cohen SN (2004) Disparate effects of MAPK mitogenic signaling. Genes Dev 12: 3008–3019. telomere attrition on gene expression during replicative senescence of human 12. Bartkova J, Rezaei N, Liontos M, Karakaidos P, Kletsas D, et al. (2006) mammary epithelial cells cultured under different conditions. 23 (37): Oncogene-induced senescence is part of the tumorigenesis barrier imposed by 6193–8. DNA damage checkpoints. Nature 444: 633–637. 27. Schwarze SR, DePrimo SE, Grabert LM, Fu VX, Brooks JD, et al. (2002) Novel 13. Poele Te RH, Okorokov AL, Jardine L, Cunnings J, Joel SP (2002) DNA pathways associated with bypassing cellular senescence in human prostate damage is able to induce senescence in tumor cells in vitro and in vivo. Cancer epithelial cells. J Biol Chem 277 (17): 14877–83. Res 62: 1876–83. 28. Johung K, Goodwin EC, Di Maio D (2007) Human papillomavirus E7 14. Faraonio R, Pane F, Intrieri M, Russo T, Cimino F (2002) In vitro acquired repression in cervical carcinoma cells initiates transcriptional cascade driven by cellular senescence and aging-specific phenotype can be distinguished on the the retinoblastoma family, resulting in senescence. J Virol 81 (5): 2102–16. basis of specific mRNA expression. Cell Death Differ 9: 862–864. 29. Oh HK, Tan AL-K, Das K, Ooi CH, Deng NT, et al. (2011) Genomic loss of 15. Parrinello S, Samper E, Krtolica A, Goldstein J, Melov S, et al. (2003) Oxygen miR-486 regulates tumor progression and the OLFM4 antiapoptotic factor in sensitivity severely limits the replicative lifespan of murine fibroblasts. Nat Cell gastric cancer. Clin Cancer Res 17: 2657–2667. Biol 5: 741–747. 30. Dimri GP, Lee X, Basile G, Acosta M, Scott G, et al. (1995) A biomarker that 16. Bartel DP (2009) MicroRNAs: target recognition and regulatory functions. Cell identifies senescent human cells in culture and in aging skin in vivo. Proc Natl 136: 215–33. Acad Sci U S A 92: 9363–9367.

PLOS ONE | www.plosone.org 9 June 2014 | Volume 9 | Issue 6 | e98669 Novel SAmiR Direct Targets

31. Diakos C, Zhong S, Xiao Y, Zhou M, Vasconcelos GM, et al. (2010) TEL- 37. Qian J, Lesage B, Beullens M, Van Eynde A, Bollen M (2011) PP1/Repo-man AML1 regulation of survivin and apoptosis via miRNA-494 and miRNA-320a. dephosphorylates mitotic histone H3 at T3 and regulates chromosomal aurora B Blood 116 (23): 4885–93. targeting. Curr Biol 21: 766–773. 32. Visvanathan J, Lee S, Lee B, Lee JW, Lee SK (2007) The microRNA miR-124 38. Vagnarelli P, Ribeiro S, Sennels L, Sanchez-Pulido L, de Lima Alves F, et al. antagonizes the anti-neural REST/SCP1 pathway during embryonic CNS (2011) Repo-Man Coordinates Chromosomal Reorganization with Nuclear development. Genes Dev 21: 744–749. Envelope Reassembly during Mitotic Exit. Dev Cell 21: 328–342. 33. Ziegelbauer JM, Sullivan CS, Ganem D (2009) Tandem array-based expression 39. Uchida F, Uzawa K, Kasamatsu A, Takatori H, Sakamoto Y, et al. (2013) screens identify host mRNA targets of virus-encoded microRNAs. Nat Genet 41: Overexpression of CDCA2 in Human Squamous Cell Carcinoma: Correlation 130–134. with Prevention of G1 Phase Arrest and Apoptosis. PLoS ONE 8 (2): e56381. 34. Peng A, Lewellyn AL, Schiemann WP, Maller JL (2010) Repo-Man controls a 40. Massari ME, Murre C (2000) Helix-loop-helix proteins: regulators of protein phosphatase 1-dependent threshold for DNA damage checkpoint transcription in eucaryotic organisms. Mol Cell Biol 20: 429–440. activation. Curr Biol 20 (5): 387–396. 41. Ruzinova MB, Benezra R (2003) Id proteins in development, cell cycle and 35. Trinkle-Mulcahy L, Andersen J, Lam YW, Moorhead G, Mann M, et al. (2006) cancer. Trends Cell Biol 13: 410–418. Repo-Man recruits PP1 gamma to chromatin and is essential for cell viability. 42. Iavarone A, Lasorella A (2006) ID proteins as targets in cancer and tools in J Cell Biol 172: 679–692. neurobiology. Trends Mol Med 12: 588–594. 36. Vagnarelli P, Hudson DF, Ribeiro SA, Trinkle-Mulcahy L, Spence JM, et al. 43. Carey JP, Knowell AE, Chinaranagari S, Chaudhary J (2013) ID4 promotes (2006) Condensin and Repo-Man-PP1 co-operate in the regulation of senescence and sensitivity to doxorubicin-induced apoptosis in DU145 prostate architecture during . Nat Cell Biol 8: 1133–1142. cancer cells. Anticancer Res 33 (10): 4271–8.

PLOS ONE | www.plosone.org 10 June 2014 | Volume 9 | Issue 6 | e98669